Public healthcare efficiency in India: bridging DEA and machine learning for effective resource allocation
Anil Gurjar, Anupam GhoshPurpose
Amid rising healthcare demands, measuring healthcare system efficiency is crucial for optimizing resource utilization and improving patient care while effectively managing healthcare costs. This study evaluates the performance of small, medium and large-capacity district hospitals (DHs) in India, and the variables that contribute to the efficiency of DHs are highlighted.
Design/methodology/approach
In the first stage, conventional data envelopment analysis (CNDEA) and slack-based measure DEA (SBM-DEA) models are employed to compute technical efficiency (TE) and identify slack in input and output variables across hospital categories. In the second stage, a unified SBM-DEA model is applied across all DHs to obtain TE scores, which are then analyzed using a random forest (RF) regression model. The RF model identifies key inputs, outputs and contextual variables – including district population (DP) and hospital size – that significantly influence TE.
Findings
Results reveal notable efficiency variation across hospital sizes, with large DHs generally performing better. Significant slack exists in human resource inputs and service availability, particularly in small and medium hospitals. RF analysis highlights the number of doctors, paramedical staff and availability of diagnostic services as key predictors of efficiency.
Originality/value
This study is among the first to evaluate Indian DHs nationwide using both DEA models and machine learning techniques. By integrating hospital size, infrastructure and environmental factors, it offers a comprehensive and scalable framework for evidence-based resource allocation and performance improvement in public healthcare.